[petsc-users] Reaching limit number of communicator with Spectrum MPI

[Victor Eijkhout]

On , 2021Aug20, at 00:33, Barry Smith <bsmith at petsc.dev<mailto:bsmith at petsc.dev>> wrote:

 It sounds like maybe the Spectrum MPI_Comm_free() is not returning the comm to the "pool" as available for future use;

1. I can not find in the standard what the proper response is to running out of communicators.

2. Mpich on my laptop returns MPI_COMM_NULL after 2044 dups without free.

3. A million dups with free run in a second.

2b. Spectrum MPI on my P9 runs out after 4096 dups

3b. Dup and free million times takes more time than writing this email. Why do I keep hearing that OpenMPI is so great? Everything slightly non-standard I try is hopelessly slow and broken.

Victor.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20210821/7a226643/attachment.html>